GRIP2: A Robust and Powerful Deep Knockoff Method for Feature Selection

2026-01-30

AI Summary

This work addresses the challenge of achieving both effective feature selection and rigorous false discovery rate (FDR) control in high-correlation, low-signal-to-noise settings—a regime where existing methods often fail. The authors propose GRIP2, a novel approach that introduces an integral framework over a two-dimensional regularization path. By integrating the activity of first-layer features from a deep network across this surface, GRIP2 constructs a naturally antisymmetric feature importance statistic. Coupled with block random sampling, this enables efficient approximation from a single model training run. The method provides finite-sample FDR guarantees and substantially outperforms current deep feature selection techniques. When applied to real-world HIV drug resistance data, GRIP2 successfully recovers known resistance-associated mutations and surpasses classical linear baselines in performance.

Technology Category

• Machine Learning: Dimensionality Reduction/Feature Selection
• Search and Optimization: Sampling/Simulation-based Search
• Computer Vision: Other Foundations of Computer Vision

Application Category

• Graph Algorithms and Modeling for the Web: Graph neural networks and deep learning approaches for Web-related graphs
• Search and Retrieval-Augmented AI: Web learning to rank, online learning, and counterfactual learning for ranking
• Web Mining and Content Analysis: Robustness and generalizability of Web mining methods

Abstract

Identifying truly predictive covariates while strictly controlling false discoveries remains a fundamental challenge in nonlinear, highly correlated, and low signal-to-noise regimes, where deep learning based feature selection methods are most attractive. We propose Group Regularization Importance Persistence in 2 Dimensions (GRIP2), a deep knockoff feature importance statistic that integrates first-layer feature activity over a two-dimensional regularization surface controlling both sparsity strength and sparsification geometry. To approximate this surface integral in a single training run, we introduce efficient block-stochastic sampling, which aggregates feature activity magnitudes across diverse regularization regimes along the optimization trajectory. The resulting statistics are antisymmetric by construction, ensuring finite-sample FDR control. In extensive experiments on synthetic and semi-real data, GRIP2 demonstrates improved robustness to feature correlation and noise level: in high correlation and low signal-to-noise ratio regimes where standard deep learning based feature selectors may struggle, our method retains high power and stability. Finally, on real-world HIV drug resistance data, GRIP2 recovers known resistance-associated mutations with power better than established linear baselines, confirming its reliability in practice.

Problem

• feature selection
• false discovery rate
• highly correlated features
• low signal-to-noise ratio
• nonlinear models

Innovation

• deep knockoff
• feature selection
• FDR control
• regularization geometry
• block-stochastic sampling